trained_model这个是什么东西？

[1015904089]

FileNotFoundError: [Errno 2] No such file or directory: 'trained_model/cloth_simulation.pth.tar' trained_model/cloth_simulation.pth.tar以及下面的一些同样在trained_model目录下的文件在哪里可以下载？还有如果要用自己的数据，需要对数据做什么样的处理？

[xusongxuliangwangsulong]

+1 How to preprocess our own data

[wangkangkan]

FileNotFoundError: [Errno 2] No such file or directory: 'trained_model/cloth_simulation.pth.tar' trained_model/cloth_simulation.pth.tar以及下面的一些同样在trained_model目录下的文件在哪里可以下载？还有如果要用自己的数据，需要对数据做什么样的处理？

+1 How to preprocess our own data

In the trained_model folder, there are the pre-trained model of the clothing simulation network and the pre-trained models of the occupancy networks representing the human body and clothing geometry. The details of this folder are as follows:

trained_model
├── cloth
│   └── latest.pth    // Pre-trained model of the clothing occupancy network
├── occ
│   └── desmpl
│         └── latest.pth // Pre-trained model of the human body occupancy network
└── cloth_simulation.pth.tar  // Pre-trained model of the clothing simulation network

For the occupancy network, we first use a single frame of the multi-view videos for training the occupancy network with all loss terms. Then we fix the occupancy model and train the other networks.

For clothing simulation network of each type of clothing, please refer to [TailorNet](https://github.com/chaitanya100100/TailorNet) for the training process and data processing procedure.

To train your own model, you should have both human body and clothing templates, and their deformation graphs for nonrigid deformation. Please refer to [Embedded deformation for shape manipulation](https://github.com/rFalque/embedded_deformation) for more details about deformation graph.

The details of the data folder are as follows:

$ROOT
├── ...
├── tools 
│   └── data   // Folder for data files
│       └── dataset // Dataset
│           ├── smpl_params // SMPL parameters for each frame
│           │   ├── 0.npy 
│           │   ├── 1.npy 
│           │   └── ... 
│           ├── templatedeformT // Files related to human and clothing templates
│           │   ├── desmpl // Files related to human template
│           │   │   ├── desmplvt.txt // Vertex positions in T-pose
│           │   │   ├── desmpltri.txt // faces
│           │   │   ├── desmpl_vfidx.txt //Face index each vertex belongs to
│           │   │   ├── nodevidx.txt // Vertex index of deformation graph nodes in the template
│           │   │   ├── modelnodeedge.txt // Edges of the deformation graph
│           │   │   ├── modelvert_node.txt // Influenced node index for each vertex
│           │   │   └── modelvert_nodeweight.txt // Weights of influenced nodes for each vertex
│           │   └── cloth // Files related to clothing template
│           │       ├── clothes_vert.txt // Vertex positions in T-pose
│           │       ├── clothes_face.txt // Template faces in T pose
│           │       ├── clothes_watertight_face.txt // Watertight model faces
│           │       ├── cloth_vfidx.txt // Face index each vertex belongs to
│           │       └── attachidx.txt // Indices of fixed vertices
│           ├── vertices // SMPL template vertex positions for each frame
│           │   ├── 0.npy 
│           │   ├── 1.npy 
│           │   └── ... 
│           ├── annots.npy // Annotations for camera parameters and image paths
│           ├── bw.npy // SMPL deformation weights
│           ├── parents.npy // SMPL parent nodes
│           └── joints.npy // SMPL skeletal joints
└── ...